Semi-interesting raw color record. S1R, A7R WB problem?

[Andrea B.]

The following raw color panels were part of some other work I was doing.

 

I white-balanced against a 5x5" Spectralon rectangle. And then took a photo of it for a white-balance reference because sometimes the camera does not measure WB quite accurately when using our strange set of dark and colourful filters.

 

Anyway, for my converted Panasonic S1R the raw color of the Spectralon was the same for both the B+W 093 IR-pass filter and a U340 stack. This struck me as being somewhat weird! Here is that color at an 80% brightness and an 80% saturation. 

Process: The raw color files from Raw Digger were subsequently averaged in PS Elements, sampled and then boosted to the 80% levels (an arbitrary choice).

 

 

 

I was also amused by the fact that the raw color of the U340 was at 341° on the color wheel. 

 

For reference's sake, here is a boosted raw color panel for the BaaderU which has a peak at 350 nm. 

That peak is near enough to the UV region which is recorded as violet or blue or blue-violet in our cameras that you can see some blue beginning to creep into the raw color.

 

 

None of this means anything very significant. Just having fun with raw colors. 

[Stefano]

So you can switch between those filters without re-doing the white balance, which is convenient.

[Andrea B.]

But this S1R is not properly white balancing. Look at this one.

[Andy Perrin]

Weiiiird! Is this WB in the camera? If so, what most likely happened is that the camera software has lower/upper limits on the color temperature and in both the UV and IR cases you are hitting them. Therefore you get the same WB in both instances. 
 

It’s very much like the incorrect WB we see when using Adobe Camera RAW or other white balance software that set arbitrary limits on the color temperature and tint. 

[Andrea B.]

The in-camera WB measured by the S1R in the visible case is quite acceptable. Attached below is a visible result made at the same time as the preceding examples. It reads as having a 1% saturation. I suppose that in an ideal world, that saturation would be 0%. But I almost always see one or two of the rgb coordinates "off" by 1 or 2 even when making WB against a Lambertian material. Nothing's perfect, after all, including reflected light and WB measurement technology. 

 

But that 6% saturation for the U340 stack is too much. And like you said, it probably indicates that I bumped up against some pre-set limit the S1R's WB software. 

 

 

 

[Andrea B.]

It's not just the S1R which has a bit of error measuring WB. Here's a WB measurement made by my old Sony A1R. It is slightly off also.

Oh well. This is no big deal, really, as it is easily corrected during conversion. I'm certainly used to that after all those years with the converted Nikons which are clueless about WB measurements in anything but visible light.

(I typo-ed the model. It is an A7R.)

 

[Andrea B.]

OR:  Maybe I should be asking myself what am I doing wrong that I can't get any of these cool cams to white balance better ??? 

[Andrea B.]

Do these in-camera white balance problems happen to anybody else when using a UV-pass filter with any of the cameras which are supposed to be able to set white balance?? Why do I never get a neutral value?? 

[Andy Perrin]

Not me? So far the Sony cameras all seemed to set the white balance in camera about the same as PhotoNinja (for example). 

[dabateman]

Olympus generally has been great.

 

Fuji, I don't have my 50S as full spectrum,  and need to test more. It can see upper 360s and under 800nm as stock.

 

My Canon S5 full spectrum compact camera didn't have issues.

 

Sigma Sd14 is monochrome for UV and IR. So its a bit different. 

 

That I think are all my cameras.

 

Because you know what to look for and can test it, might just be the reason you have issues Andrea  

[colinbm]

White Balance is by name only, it can be grey or black too.
You can in-camera WB off reflected light from a PTFE target or defused light with a thin, sub millimeter film of PTFE in front of the lens.

[Andy Perrin]

Strictly speaking, it should be 18% reflectivity if you want to do things analogous to what a Color Checker does in visible light. The 18% reflectivity equates to 50% gray if you have gamma of 2.4. It’s a good standard because white balancing off a black object gives you mostly noise (in the absence of light, only noise remains) and white balancing off a bright enough white will saturate the sensor. So 50% grey or 18% reflectivity is a good intermediate value. 
 

Of course we rarely need such precision! But it is still a good idea to pick something intermediate to wb with. 

[ulf]

That is a good approach.

When I did my flash illuminated UV photography I placed my matted PTFE reference further away from the flash, but still in the image to get a less bright area for the WB.

For sun-illuminated scenes you could instead take a separate image of the reference surface with shorter exposure time, before doing the final UV-shoots.

 

As I mainly shoot RAW I have very little experience with in camera WB and have nothing to add there.

 

 

 

[lukaszgryglicki]

My Nikon D600 is not able to whitebalance at all, another D600 is mono - so it makes no sense to wb it. Fuji 50R quartz will also be mono...

I will eventually do 3 photos with different filters and use them as channels for RGB image.

I also bought Accu Raw Monochrome, so I will be taking advantage of DeBayering - it makes photos from D600 mono sharper - expect the same for 50Rqm (quartz mono).